CN111142027A - Lithium iron phosphate battery state-of-charge monitoring and early warning method based on neural network - Google Patents

Lithium iron phosphate battery state-of-charge monitoring and early warning method based on neural network Download PDF

Info

Publication number
CN111142027A
CN111142027A CN201911423345.7A CN201911423345A CN111142027A CN 111142027 A CN111142027 A CN 111142027A CN 201911423345 A CN201911423345 A CN 201911423345A CN 111142027 A CN111142027 A CN 111142027A
Authority
CN
China
Prior art keywords
fuzzy
battery
neural network
early warning
soc
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201911423345.7A
Other languages
Chinese (zh)
Inventor
陈良亮
张�浩
张卫国
周静
周材
邵军军
孙季泽
陈嘉栋
余洋
杨凤坤
赵明宇
孙广明
仇新宇
李波
许庆强
崔文佳
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
State Grid Jiangsu Electric Power Co ltd Marketing Service Center
State Grid Corp of China SGCC
State Grid Jiangsu Electric Power Co Ltd
NARI Group Corp
Nari Technology Co Ltd
NARI Nanjing Control System Co Ltd
Electric Power Research Institute of State Grid Jiangsu Electric Power Co Ltd
State Grid Electric Power Research Institute
Suzhou Power Supply Co of State Grid Jiangsu Electric Power Co Ltd
Original Assignee
State Grid Corp of China SGCC
State Grid Jiangsu Electric Power Co Ltd
NARI Group Corp
Nari Technology Co Ltd
NARI Nanjing Control System Co Ltd
Electric Power Research Institute of State Grid Jiangsu Electric Power Co Ltd
State Grid Electric Power Research Institute
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by State Grid Corp of China SGCC, State Grid Jiangsu Electric Power Co Ltd, NARI Group Corp, Nari Technology Co Ltd, NARI Nanjing Control System Co Ltd, Electric Power Research Institute of State Grid Jiangsu Electric Power Co Ltd, State Grid Electric Power Research Institute filed Critical State Grid Corp of China SGCC
Priority to CN201911423345.7A priority Critical patent/CN111142027A/en
Publication of CN111142027A publication Critical patent/CN111142027A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/367—Software therefor, e.g. for battery testing using modelling or look-up tables
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/382—Arrangements for monitoring battery or accumulator variables, e.g. SoC
    • G01R31/3842—Arrangements for monitoring battery or accumulator variables, e.g. SoC combining voltage and current measurements

Abstract

The invention discloses a lithium iron phosphate battery state of charge monitoring and early warning method based on a neural network, which is characterized in that a T-S fuzzy neural network model is trained by utilizing a training data set to obtain a trained T-S fuzzy neural network model; inputting actual battery voltage, battery current and battery temperature data at the T moment by the trained T-S fuzzy neural network model, and outputting the estimated battery SOC; when the SOC reduction rate of the battery is higher than a set value, an early warning signal that the electric quantity is reduced too fast and the battery is abnormal is sent out; and when the SOC of the battery is lower than 10%, sending out an early warning signal of low battery. The lithium iron phosphate battery state-of-charge monitoring and early warning method based on the neural network can effectively improve SOC estimation precision, monitor and early warn the battery state-of-charge, and improve the reliability and safety of the battery.

Description

Lithium iron phosphate battery state-of-charge monitoring and early warning method based on neural network
Technical Field
The invention relates to a lithium iron phosphate battery state of charge monitoring and early warning method based on a neural network, and belongs to the technical field of battery state of charge monitoring and early warning.
Background
Currently, in order to deal with energy crisis and reduce global warming, many countries are beginning to pay attention to emission reduction, energy conservation and low-carbon economy development. Because the electric automobile can reduce the emission of carbon dioxide by using electric drive, even realize zero emission, the electric automobile draws attention from various countries and develops rapidly. With the development of electric vehicles, more and more electric vehicles use lithium batteries as power sources to cut off, but the cost of the batteries is still high, and the performance and price of the power batteries become the main bottleneck of the development of the electric vehicles.
The lithium iron phosphate battery has long service life, good safety performance and low cost, and becomes an ideal power source for the electric automobile. SOC is one of the key parameters that indicate battery utilization. However, since the batteries are connected in series and parallel, the safety of charge and discharge is considered, and thus, direct measurement cannot be performed.
With the development of electric vehicles, Battery Management Systems (BMS) have been increasingly applied. In order to fully exert the dynamic performance of the battery system, increase the use safety of the battery system, prevent the charging and discharging of the battery, prolong the service life of the battery, optimize the driving performance and improve the performance of the electric automobile, the application program uses the SOC of the battery of the BMS system to accurately estimate the SOC. Estimation of the SOC of an electric vehicle is the basis of a battery management system. Improving the estimation accuracy has important significance for improving the service efficiency, prolonging the service life of the battery, improving the reliability of the battery and improving the safety.
In the past strategy, because the current measurement is inaccurate, accumulated errors exist, the open-circuit voltage method can accurately estimate the SOC of the battery, but the open-circuit voltage can only be measured for a long time after the battery stops working, and the method is not suitable for online estimation.
Disclosure of Invention
The purpose is as follows: in order to overcome the defects in the prior art and better monitor the state of charge of the lithium iron phosphate battery, the invention provides a neural network-based lithium iron phosphate battery state of charge monitoring and early warning method. If too many factors are selected, the calculation will be very heavy. If too few factors are selected, the artificial neural network cannot correctly reflect the SOC. The invention combines the fuzzy mathematical principle, establishes the T-S fuzzy neural network, reasonably selects the factor boundary, can effectively estimate the SOC, improves the estimation precision and has higher calculation efficiency.
The technical scheme is as follows: in order to solve the technical problems, the technical scheme adopted by the invention is as follows:
a lithium iron phosphate battery state of charge monitoring and early warning method based on a neural network comprises the following steps:
training the T-S fuzzy neural network model by utilizing a training data set to obtain a trained T-S fuzzy neural network model;
and inputting actual battery voltage, battery current and battery temperature data at the T moment by the trained T-S fuzzy neural network model, and outputting the estimated battery SOC.
Preferably, the method further comprises the following steps:
when the SOC reduction rate of the battery is higher than a set value, an early warning signal that the electric quantity is reduced too fast and the battery is abnormal is sent out; and when the SOC of the battery is lower than 10%, sending out an early warning signal of low battery.
The utility model provides a lithium iron phosphate battery state of charge monitoring and early warning device based on neural network, includes following module:
a network model training module: training the T-S fuzzy neural network model by utilizing a training data set to obtain a trained T-S fuzzy neural network model;
a network model estimation module: and inputting actual battery voltage, battery current and battery temperature data at the T moment by the trained T-S fuzzy neural network model, and outputting the estimated battery SOC.
Preferably, the system also comprises the following modules:
the early warning module: when the SOC reduction rate of the battery is higher than a set value, an early warning signal that the electric quantity is reduced too fast and the battery is abnormal is sent out; and when the SOC of the battery is lower than 10%, sending out an early warning signal of low battery.
As a preferred scheme, the T-S fuzzy neural network model is trained by utilizing a training data set to obtain the trained T-S fuzzy neural network model, and the specific steps are as follows:
(1) the T-S fuzzy neural network model comprises an input layer, a fuzzy rule calculation layer and an output layer; inputting a training input quantity X ═ X formed by historical battery voltage, battery current and battery temperature data1,x2,x3]Training output expected value yi；
(2) Output expectation y through known training in the output layeriCalculating the fuzzy operator theta in the fuzzy rule calculation layeriI is the number of rules;
(3) according to the fuzzy operator thetaiCalculating fuzzy membership value in fuzzy layer
Figure BDA0002348241580000021
xj∈x1,x2,x3，
Figure BDA0002348241580000022
j belongs to 1,2 and 3 is a fuzzy set of training input quantity, and i is a rule number;
(4) according to fuzzy membership value
Figure BDA0002348241580000023
Training input quantity X ═ X1,x2,x3]Calculating the center of the function
Figure BDA0002348241580000024
Width of function
Figure BDA0002348241580000025
(5) The dimension of the input node is 3, the dimension of the output node is 1, the dimension of the hidden node is 6, the coefficient learning rate α is set as a random constant, the parameter learning rate β is set as a random constant, and the iteration number G of the algorithmmaxSet to an integer less than 100.
Preferably, the battery SOC decrease rate k is 100 × SOC (T +1) -SOC (T) -l, T representing the tth time, and the set value is set to 5.
Preferably, the training output expected value yiThe calculation formula of (a) is as follows:
Figure BDA0002348241580000026
θiin order to perform the fuzzy operator,
Figure BDA0002348241580000027
for the model real-valued parameter, the input quantity X ═ X1,x2,...,xr]I is the number of rules, n is the total number of rules;
fuzzy operator thetaiThe calculation formula of (a) is as follows:
Figure BDA0002348241580000031
input quantity X ═ X1,x2,...,xr]Fuzzy membership value
Figure BDA0002348241580000032
Fuzzy membership value
Figure BDA0002348241580000033
The calculation formula of (a) is as follows:
Figure BDA0002348241580000034
Figure BDA0002348241580000035
as the center of the function is taken,
Figure BDA0002348241580000036
for the width of the function, input quantity xj，
Figure BDA0002348241580000037
J is a fuzzy set of input quantities, and j is 0,1, 2.
Has the advantages that: the lithium iron phosphate battery state-of-charge monitoring and early warning method based on the neural network can effectively improve SOC estimation precision, monitor and early warn the battery state-of-charge, and improve the reliability and safety of the battery.
Drawings
FIG. 1 is a flow chart of a warning method of the present invention;
FIG. 2 is a T-S fuzzy neural network monitoring SOC diagram;
FIG. 3 is a T-S fuzzy neural network monitoring SOC error graph;
FIG. 4 is a schematic view of the structure of the device of the present invention.
Detailed Description
The present invention will be further described with reference to the accompanying drawings.
As shown in fig. 1, a method for monitoring and early warning the state of charge of a lithium iron phosphate battery based on a neural network comprises the following steps:
and (3) applying the T-S fuzzy neural network algorithm to model training, and applying the training model to SOC estimation of the lithium iron phosphate battery.
Step 1, firstly, a T-S fuzzy neural network regression model is required to be established, wherein parameters of the regression model comprise input node dimension, output node dimension, hidden node dimension, coefficient learning rate α, parameter learning rate β and iteration times G of the algorithmmaxFunction center c, initialization of width b.
And 2, constructing a model training data set according to the NASA database, taking the battery voltage, the battery current and the battery temperature as input quantities of the model training data set, and taking the corresponding SOC as an output quantity of the model training data set.
And 3, executing a T-S fuzzy neural network algorithm for model training, and forming an SOC monitoring model along with continuous updating of network parameters in the training process.
And 4, when the SOC monitoring model is used for monitoring the SOC in real time, monitoring the SOC of the battery according to the real-time current and voltage of the battery and the temperature time sequence data of the battery.
Step 5, monitoring the SOC in real time and calculating a reduction rate, and warning that the reduction rate is too fast and the battery is abnormal when the reduction rate is higher than a set value; and when the SOC is lower than 10%, a low-power early warning signal is sent out, so that the monitoring and early warning of the SOC of the battery are realized.
Firstly, a T-S fuzzy neural network regression model is constructed according to a T-S fuzzy algorithm. The T-S fuzzy neural network regression model is mainly divided into the following four parts: the input layer, the fuzzy rule calculation layer and the output layer.
The SOC monitoring model obtained by training the T-S fuzzy neural network regression model established by the invention is mainly applied to SOC state monitoring of the lithium iron phosphate battery, the input quantity of the SOC monitoring model comprises three data of battery voltage, battery current and battery temperature, and the output quantity of the SOC monitoring model is corresponding SOC.
The SOC monitoring model trained by the invention can monitor the battery state in real time and set two early warning mechanisms of a decline rate early warning and a low power early warning.
T-S fuzzy algorithm
The T-S fuzzy system has strong self-adaptive capacityThe fuzzy subsets of the membership functions may be automatically updated and corrected. The T-S fuzzy system may be defined according to the "if-then" rule, at RiUnder control of the rules, the control rules have the following form:
Figure BDA0002348241580000041
Figure BDA0002348241580000042
wherein i is the number of rules,
Figure BDA0002348241580000043
is a fuzzy set of T-S fuzzy system input quantities,
Figure BDA0002348241580000044
is a model real-valued parameter, yiIs the system output according to the ith fuzzy rule. I.e. a part of the input is ambiguous and a part of the output is deterministic.
Suppose there is an input of X ═ X1,x2,...,xr]First, each input variable x is calculated according to a fuzzy rulekDegree of membership. The invention selects a Gaussian function as a membership function of a T-S fuzzy system, and the membership function is as follows:
Figure BDA0002348241580000045
wherein
Figure BDA0002348241580000046
And
Figure BDA0002348241580000047
is an adjustable parameter of the membership function,
Figure BDA0002348241580000048
is the center of the gaussian function and,
Figure BDA0002348241580000049
is the width of the gaussian function.
Fuzzy operator theta in T-S fuzzy systemiDefined as the following equation:
Figure BDA00023482415800000410
according to the calculation result of the T-S fuzzy system, the output value y of the T-S fuzzy system can be calculatediComprises the following steps:
Figure BDA00023482415800000411
T-S fuzzy neural network model
And constructing a T-S fuzzy neural network model according to the T-S fuzzy algorithm. The T-S fuzzy neural network model is divided into the following four parts: the fuzzy rule calculation method comprises an input layer, a fuzzy rule calculation layer and an output layer. The mutual logical relationship constructed by the network model is as follows:
the input layer is connected to the input vector and the number of nodes is equal to the dimension of the input vector.
In the fuzzy layer, a fuzzy membership value mu is calculated by a membership function formula (3).
In the calculation layer of the fuzzy rule, the function formula (4) calculates a fuzzy operator θ.
In the output layer, the output y of the T-S fuzzy neural network model calculated by the formula (5)i。
The T-S fuzzy neural network model training algorithm can be realized by the following strategies:
input calculation error η:
Figure BDA0002348241580000051
in formula (6), ydIs the output of the expected value of the network model, ycIs the output of the actual values of the network model, η is ydAnd ycThe calculation error therebetween.
And (3) correcting the coefficient:
Figure BDA0002348241580000052
the normalized total objective function is:
Figure BDA0002348241580000053
in the formulae (7) and (8),
Figure BDA0002348241580000054
is the coefficient of the T-S fuzzy neural network, α is the coefficient learning rate, xiIs an input parameter to the network. ThetaiAnd k is a fuzzy operator formed by multiplying membership degrees, j belongs to (1, 2.. eta., r), and i is a regular number.
According to the gradient descent method, in the process of network learning, the precondition parameters are corrected, and the parameter correction quantity is as follows:
Figure BDA0002348241580000055
Figure BDA0002348241580000056
in the formulae (9) and (10),
Figure BDA0002348241580000057
and
Figure BDA0002348241580000058
is an adjustable parameter of the membership function,
Figure BDA0002348241580000059
is the center of the gaussian function and,
Figure BDA00023482415800000510
for the width of the gaussian function, β is the parameter learning rate, k ∈ j, j ∈ (1, 2.. eta., r), i is the number of rules.
SOC monitoring and early warning algorithm process based on fuzzy neural network
In the T-S fuzzy neural network model, the number of output and input nodes and a fuzzy membership function are determined by the dimension of a training sample. In the invention, the battery voltage, the battery current and the battery temperature of a storage battery are taken as training inputs of a model, and the corresponding SOC is taken as a training output of the model, so the dimension of input data is 3 dimensions, and the dimension of output data is 1 dimension.
Estimating the SOC by adopting a T-S fuzzy neural network model, which comprises the following specific steps:
1. generating SOC monitoring model based on T-S fuzzy neural network
(1) Inputting historical battery voltage, battery current and battery temperature data to form X ═ X1,x2,x3]Network output expected value ytest。
(2) The method comprises the steps of inputting the dimension of nodes, outputting the dimension of nodes, hiding the dimension of nodes, learning coefficient α, learning parameter β and iteration times G of the algorithmmaxMembership function center c, width b.
(3) Input data is normalized.
(4) Calculating the fuzzy membership value mu, the fuzzy operator theta and the network training actual value y of the networktrain。
(5) Updating coefficients
Figure BDA0002348241580000061
(6) Judging the end condition, if the number of iterations reaches GmaxAnd ending the algorithm; otherwise, returning to the step 4.
2. Real-time monitoring and early warning SOC
(7) And generating a fuzzy neural network model, loading actual battery voltage, battery current and battery temperature data at the time T in real time, and outputting a network estimation SOC (T).
(8) Monitoring the SOC once every 1 unit time, simultaneously calculating a descending rate k of 100 SOC (T +1) -SOC (T), and if k is larger than 5, early warning that the descending rate is too high and the battery is abnormal. And meanwhile, if the SOC is less than 10%, early warning that the battery power is low.
The experimental data of the stable discharge process of the lithium iron phosphate battery in the NASA Ames database is used as a training data set, wherein a group of four batteries (No. 5, No. 6, No. 7 and No. 18) of the lithium iron phosphate battery work at room temperature through three different working characteristics (a charging mode, a discharging mode and an impedance mode). Charging in constant current mode of 1.5A until the battery voltage reaches 4.2V, and then charging in constant current mode until the charging current drops to 20 mA. The discharge was performed at a constant current level of 2A until the cell voltage dropped to 2.7V, 2.5V, 2.2V and 2.5V. The impedance in the range of 0.1Hz to 5khz was measured by electrochemical impedance spectroscopy. Repeated charge and discharge cycles lead to accelerated aging of the battery, and impedance measurements can make the insight that internal parameters of the battery change as aging progresses. When the battery reached the end of service (EOL) standard, the experiment was stopped and the rated capacity was reduced by 30% (from 2Ahr to 1.4 Ahr). This data set can be used to monitor remaining power, and the specific embodiment of the present invention uses battery data No. 5 as a training data set for 500 groups of data.
This data set can be used to monitor the remaining charge (for a given discharge cycle) and the Remaining Useful Life (RUL). Discharge data for battery No. 5 is used as the experimental data source. Since the entire discharge process is recorded, the actual SOC can be calculated by equation (11).
Figure BDA0002348241580000062
In the formula (11), t is a discharge time when the discharge voltage reaches the cut-off voltage, IdcIs the discharge current.
In order to better reflect the accuracy of the algorithm, a unified training set is used for training and charge SOC monitoring is carried out. During discharge, with the battery parameters (voltage, current and temperature) as model inputs and the corresponding SOC as model outputs, this change is mainly compared to previous actual SOC sampling points to evaluate the performance of the T-S fuzzy neural network regression.
The monitoring model based on the T-S fuzzy neural network is tested by 100 groups of unit time sequence data for testing the network generalization capability, and partial test results are shown in Table 1 and comprise five groups of data of voltage, current, temperature, actually measured SOC and monitored SOC.
TABLE 1 SOC monitoring results based on fuzzy neural network
voltage/V current/A Temperature/. degree.C Actual SOC Monitoring SOC
4.23 2.010 25.1 0.90 0.92
4.01 2.010 25.6 0.89 0.90
3.99 2.009 26.1 0.88 0.90
3.91 2.009 26.5 0.87 0.89
3.90 2.008 27.0 0.87 0.86
3.88 2.007 27.4 0.86 0.85
3.85 2.006 27.9 0.85 0.84
3.83 2.006 28.3 0.81 0.80
3.82 2.006 28.7 0.83 0.82
3.80 2.006 29.1 0.80 0.79
3.78 2.006 29.6 0.79 0.80
3.77 2.005 30.1 0.77 0.75
3.76 2.005 30.4 0.78 0.77
3.75 2.004 30.7 0.69 0.68
3.74 2.004 31.1 0.71 0.70
3.72 2.002 31.4 0.70 0.72
3.70 2.001 31.7 0.68 0.67
3.69 2.000 31.9 0.63 0.61
In the test data set, 100 groups of unit time sequence data are selected for testing the network, and 100 groups of data are imported into a monitoring model for simulation analysis. The results of 100 sets of data tests are shown in FIG. 2, and the error results are shown in FIG. 3.
As the SOC changes, the monitoring and warning signals of the model are generated and summed as shown in table 2: and early warning of overhigh drop rate is generated in unit time 23 and unit time 27, and early warning of low battery is continuously carried out in 80-100 periods.
TABLE 2 SOC monitoring and early warning signal
Early warning signal Unit time T
Early warning of too high rate of decline 23
Early warning of too high rate of decline 27
Low battery warning 80-100
Experimental results show that the output of the SOC monitoring model based on the T-S fuzzy neural network can better track the actual SOC of the storage battery, the model has higher estimation precision, in order to visually reflect the monitoring precision of the monitoring model, a sample is represented by an abscissa in the graph 3, and a monitoring error of the model is represented by an ordinate. Through simulation calculation, the maximum value of the error is 0.029, and the monitoring model based on the T-S fuzzy neural network has good generalization capability. Finally, the model can monitor the SOC in real time, two early warning mechanisms of over-high decline rate early warning and low power early warning are provided, and the SOC real-time monitoring early warning is realized.
As shown in fig. 4, a lithium iron phosphate battery state of charge monitoring and early warning device based on a neural network comprises the following modules:
a network model training module: training the T-S fuzzy neural network model by utilizing a training data set to obtain a trained T-S fuzzy neural network model;
a network model estimation module: inputting actual battery voltage, battery current and battery temperature data at the T moment by the trained T-S fuzzy neural network model, and outputting the estimated battery SOC;
the early warning module: when the SOC reduction rate of the battery is higher than a set value, an early warning signal that the electric quantity is reduced too fast and the battery is abnormal is sent out; and when the SOC of the battery is lower than 10%, sending out an early warning signal of low battery.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The above description is only of the preferred embodiments of the present invention, and it should be noted that: it will be apparent to those skilled in the art that various modifications and adaptations can be made without departing from the principles of the invention and these are intended to be within the scope of the invention.

Claims (10)

1. A lithium iron phosphate battery state of charge monitoring and early warning method based on a neural network is characterized by comprising the following steps: the method comprises the following steps:
training the T-S fuzzy neural network model by utilizing a training data set to obtain a trained T-S fuzzy neural network model;
and inputting actual battery voltage, battery current and battery temperature data at the T moment by the trained T-S fuzzy neural network model, and outputting the estimated battery SOC.
2. The lithium iron phosphate battery state-of-charge monitoring and early warning method based on the neural network as claimed in claim 1, wherein: also comprises the following steps:
when the SOC reduction rate of the battery is higher than a set value, an early warning signal that the electric quantity is reduced too fast and the battery is abnormal is sent out; and when the SOC of the battery is lower than 10%, sending out an early warning signal of low battery.
3. The lithium iron phosphate battery state-of-charge monitoring and early warning method based on the neural network as claimed in claim 1 or 2, characterized in that:
training the T-S fuzzy neural network model by utilizing a training data set to obtain the trained T-S fuzzy neural network model, and specifically comprising the following steps:
(1) the T-S fuzzy neural network model comprises an input layer, a fuzzy rule calculation layer and an output layer; inputting a training input quantity X ═ X formed by historical battery voltage, battery current and battery temperature data1,x2,x3]Training output expected value yi；
(2) Output expectation y through known training in the output layeriCalculating the fuzzy operator theta in the fuzzy rule calculation layeriI is the number of rules;
(3) according to the fuzzy operator thetaiCalculating fuzzy membership value in fuzzy layer
Figure FDA0002348241570000011
xj∈x1,x2,x3，
Figure FDA0002348241570000012
Is a fuzzy set of training input quantities, i is a rule number;
(4) according to fuzzy membership value
Figure FDA0002348241570000013
Training input quantity X ═ X1,x2,x3]Calculating the center of the function
Figure FDA0002348241570000014
Width of function
Figure FDA0002348241570000015
(5) The dimension of the input node is 3, the dimension of the output node is 1, the dimension of the hidden node is 6, the coefficient learning rate α is set as a random constant, the parameter learning rate β is set as a random constant, and the iteration number G of the algorithmmaxSet to an integer less than 100.
4. The lithium iron phosphate battery state-of-charge monitoring and early warning method based on the neural network as claimed in claim 2, characterized in that: the battery SOC decrease rate k is 100 × SOC (T +1) -SOC (T) |, T representing the tth time, and the set value is set to 5.
5. The lithium iron phosphate battery state-of-charge monitoring and early warning method based on the neural network as claimed in claim 3, wherein: training output expected value yiThe calculation formula of (a) is as follows:
Figure FDA0002348241570000016
θiin order to perform the fuzzy operator,
Figure FDA0002348241570000017
for the model real-valued parameter, the input quantity X ═ X1,x2,...,xr]I is the number of rules, n is the total number of rules;
fuzzy operator thetaiThe calculation formula of (a) is as follows:
Figure FDA0002348241570000021
input quantity X ═ X1,x2,...,xr]Fuzzy membership value
Figure FDA0002348241570000022
Fuzzy membership value
Figure FDA0002348241570000023
The calculation formula of (a) is as follows:
Figure FDA0002348241570000024
Figure FDA0002348241570000025
as the center of the function is taken,
Figure FDA0002348241570000026
for the width of the function, input quantity xj，
Figure FDA0002348241570000027
J is a fuzzy set of input quantities, and j is 0,1, 2.
6. The utility model provides a lithium iron phosphate battery state of charge monitoring and early warning device based on neural network which characterized in that: the system comprises the following modules:
a network model training module: training the T-S fuzzy neural network model by utilizing a training data set to obtain a trained T-S fuzzy neural network model;
a network model estimation module: and inputting actual battery voltage, battery current and battery temperature data at the T moment by the trained T-S fuzzy neural network model, and outputting the estimated battery SOC.
7. The lithium iron phosphate battery state of charge monitoring and early warning device of the neural network of claim 6, characterized in that: the system also comprises the following modules:
the early warning module: when the SOC reduction rate of the battery is higher than a set value, an early warning signal that the electric quantity is reduced too fast and the battery is abnormal is sent out; and when the SOC of the battery is lower than 10%, sending out an early warning signal of low battery.
8. The lithium iron phosphate battery state of charge monitoring and early warning device of the neural network as claimed in claim 6 or 7, characterized in that:
training the T-S fuzzy neural network model by utilizing a training data set to obtain the trained T-S fuzzy neural network model, and specifically comprising the following steps:
(1) the T-S fuzzy neural network model comprises an input layer, a fuzzy rule calculation layer and an output layer; inputting a training input quantity X ═ X formed by historical battery voltage, battery current and battery temperature data1,x2,x3]Training output expected value yi；
(2) Output expectation y through known training in the output layeriCalculating the fuzzy operator theta in the fuzzy rule calculation layeriI is the number of rules;
(3) according to the fuzzy operator thetaiCalculating fuzzy membership value in fuzzy layer
Figure FDA0002348241570000028
xj∈x1,x2,x3，
Figure FDA0002348241570000029
Is a fuzzy set of training input quantities, i is a rule number;
(4) according to fuzzy membership value
Figure FDA00023482415700000210
Training input quantity X ═ X1,x2,x3]CalculatingCenter of function
Figure FDA00023482415700000211
Width of function
Figure FDA00023482415700000212
(5) The dimension of the input node is 3, the dimension of the output node is 1, the dimension of the hidden node is 6, the coefficient learning rate α is set as a random constant, the parameter learning rate β is set as a random constant, and the iteration number G of the algorithmmaxSet to an integer less than 100.
9. The lithium iron phosphate battery state of charge monitoring and early warning device based on the neural network as claimed in claim 7, characterized in that: the battery SOC decrease rate k is 100 × SOC (T +1) -SOC (T) |, T representing the tth time, and the set value is set to 5.
10. The lithium iron phosphate battery state of charge monitoring and early warning device based on the neural network as claimed in claim 8, characterized in that: training output expected value yiThe calculation formula of (a) is as follows:
Figure FDA0002348241570000031
θiin order to perform the fuzzy operator,
Figure FDA0002348241570000032
for the model real-valued parameter, the input quantity X ═ X1,x2,...,xr]I is the number of rules, n is the total number of rules;
fuzzy operator thetaiThe calculation formula of (a) is as follows:
Figure FDA0002348241570000033
input quantity X ═ X1,x2,...,xr]Fuzzy membership value
Figure FDA0002348241570000034
Fuzzy membership value
Figure FDA0002348241570000035
The calculation formula of (a) is as follows:
Figure FDA0002348241570000036
Figure FDA0002348241570000037
as the center of the function is taken,
Figure FDA0002348241570000038
for the width of the function, input quantity xj，
Figure FDA0002348241570000039
J is a fuzzy set of input quantities, and j is 0,1, 2.
CN201911423345.7A 2019-12-31 2019-12-31 Lithium iron phosphate battery state-of-charge monitoring and early warning method based on neural network Pending CN111142027A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911423345.7A CN111142027A (en) 2019-12-31 2019-12-31 Lithium iron phosphate battery state-of-charge monitoring and early warning method based on neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911423345.7A CN111142027A (en) 2019-12-31 2019-12-31 Lithium iron phosphate battery state-of-charge monitoring and early warning method based on neural network

Publications (1)

Publication Number Publication Date
CN111142027A true CN111142027A (en) 2020-05-12

Family

ID=70523176

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911423345.7A Pending CN111142027A (en) 2019-12-31 2019-12-31 Lithium iron phosphate battery state-of-charge monitoring and early warning method based on neural network

Country Status (1)

Country Link
CN (1) CN111142027A (en)

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111796185A (en) * 2020-06-16 2020-10-20 合肥力高动力科技有限公司 Lithium iron phosphate battery SOC-OCV calibration method based on T-S type fuzzy algorithm
CN111952962A (en) * 2020-07-30 2020-11-17 国网江苏省电力有限公司南京供电分公司 Power distribution network low voltage prediction method based on T-S fuzzy neural network
CN111983468A (en) * 2020-08-24 2020-11-24 哈尔滨理工大学 Safety degree estimation method of lithium power battery based on neural network
CN111983467A (en) * 2020-08-24 2020-11-24 哈尔滨理工大学 Battery safety degree estimation method and estimation device based on second-order RC equivalent circuit model
CN112782588A (en) * 2020-12-30 2021-05-11 深圳市加码能源科技有限公司 SOC online monitoring method based on LSSVM and storage medium thereof
CN113658415A (en) * 2021-07-30 2021-11-16 南京凡科信息科技有限公司 Early warning method and system for intelligent gateway
CN114528772A (en) * 2022-04-20 2022-05-24 深圳市森树强电子科技有限公司 Charger charging prediction method in electromechanical converter control system
WO2022183459A1 (en) * 2021-03-04 2022-09-09 宁德时代新能源科技股份有限公司 Method and apparatus for estimating soc of battery pack, and battery management system
WO2022248532A1 (en) * 2021-05-25 2022-12-01 Danmarks Tekniske Universitet Data-driven and temperature-cycles based remaining useful life estimation of an electronic device

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101359036A (en) * 2007-07-31 2009-02-04 比亚迪股份有限公司 Method for measuring state of charge of battery
CN101964018A (en) * 2010-08-26 2011-02-02 湘潭大学 SOC estimation method of battery of electric vehicle
CN106482507A (en) * 2016-10-18 2017-03-08 湖南大学 A kind of cement decomposing furnace combustion automatic control method
CN106971238A (en) * 2017-03-10 2017-07-21 东华大学 The Short-Term Load Forecasting Method of Elman neutral nets is obscured based on T S
CN107380004A (en) * 2017-09-07 2017-11-24 苏州易信安工业技术有限公司 A kind of electric vehicle battery management method, apparatus and system
CN107861070A (en) * 2017-10-25 2018-03-30 北京交通大学 A kind of health state of lithium ion battery inline diagnosis method
CN109840335A (en) * 2017-11-27 2019-06-04 上海交通大学 Based on the radial forging pit prediction optimization method for strengthening T-S fuzzy neural network
CN110310266A (en) * 2019-06-26 2019-10-08 江苏理工学院 A kind of acetes chinensis method based on T-S fuzzy neural network
CN110488202A (en) * 2019-07-24 2019-11-22 北京航空航天大学 Vehicular battery state-of-charge estimation method based on deep neural network

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101359036A (en) * 2007-07-31 2009-02-04 比亚迪股份有限公司 Method for measuring state of charge of battery
CN101964018A (en) * 2010-08-26 2011-02-02 湘潭大学 SOC estimation method of battery of electric vehicle
CN106482507A (en) * 2016-10-18 2017-03-08 湖南大学 A kind of cement decomposing furnace combustion automatic control method
CN106971238A (en) * 2017-03-10 2017-07-21 东华大学 The Short-Term Load Forecasting Method of Elman neutral nets is obscured based on T S
CN107380004A (en) * 2017-09-07 2017-11-24 苏州易信安工业技术有限公司 A kind of electric vehicle battery management method, apparatus and system
CN107861070A (en) * 2017-10-25 2018-03-30 北京交通大学 A kind of health state of lithium ion battery inline diagnosis method
CN109840335A (en) * 2017-11-27 2019-06-04 上海交通大学 Based on the radial forging pit prediction optimization method for strengthening T-S fuzzy neural network
CN110310266A (en) * 2019-06-26 2019-10-08 江苏理工学院 A kind of acetes chinensis method based on T-S fuzzy neural network
CN110488202A (en) * 2019-07-24 2019-11-22 北京航空航天大学 Vehicular battery state-of-charge estimation method based on deep neural network

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
SHUXIANG SONG 等: "State-of-charge (SOC) estimation using T-S Fuzzy Neural Network for Lithium Iron Phosphate Battery", 《2018 26TH INTERNATIONAL CONFERENCE ON SYSTEMS ENGINEERING (ICSENG)》 *
韦振汉: "锂离子电池荷电及健康状态预测方法研究", 《中国优秀硕士学位论文全文数据库 工程科技II辑》 *

Cited By (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111796185A (en) * 2020-06-16 2020-10-20 合肥力高动力科技有限公司 Lithium iron phosphate battery SOC-OCV calibration method based on T-S type fuzzy algorithm
CN111796185B (en) * 2020-06-16 2022-11-08 合肥力高动力科技有限公司 Lithium iron phosphate battery SOC-OCV calibration method based on T-S type fuzzy algorithm
CN111952962A (en) * 2020-07-30 2020-11-17 国网江苏省电力有限公司南京供电分公司 Power distribution network low voltage prediction method based on T-S fuzzy neural network
CN111983467A (en) * 2020-08-24 2020-11-24 哈尔滨理工大学 Battery safety degree estimation method and estimation device based on second-order RC equivalent circuit model
CN111983468A (en) * 2020-08-24 2020-11-24 哈尔滨理工大学 Safety degree estimation method of lithium power battery based on neural network
CN111983468B (en) * 2020-08-24 2022-11-18 哈尔滨理工大学 Safety degree estimation method of lithium power battery based on neural network
CN111983467B (en) * 2020-08-24 2023-02-03 哈尔滨理工大学 Battery safety degree estimation method and estimation device based on second-order RC equivalent circuit model
CN112782588A (en) * 2020-12-30 2021-05-11 深圳市加码能源科技有限公司 SOC online monitoring method based on LSSVM and storage medium thereof
CN112782588B (en) * 2020-12-30 2023-02-17 深圳市加码能源科技有限公司 SOC online monitoring method based on LSSVM and storage medium thereof
WO2022183459A1 (en) * 2021-03-04 2022-09-09 宁德时代新能源科技股份有限公司 Method and apparatus for estimating soc of battery pack, and battery management system
WO2022248532A1 (en) * 2021-05-25 2022-12-01 Danmarks Tekniske Universitet Data-driven and temperature-cycles based remaining useful life estimation of an electronic device
CN113658415A (en) * 2021-07-30 2021-11-16 南京凡科信息科技有限公司 Early warning method and system for intelligent gateway
CN113658415B (en) * 2021-07-30 2024-03-26 江苏湛德医疗用品有限公司 Early warning method and system of intelligent gateway
CN114528772A (en) * 2022-04-20 2022-05-24 深圳市森树强电子科技有限公司 Charger charging prediction method in electromechanical converter control system
CN114528772B (en) * 2022-04-20 2022-07-01 深圳市森树强电子科技有限公司 Charger charging prediction method in electromechanical converter control system

Similar Documents

Publication Publication Date Title
CN111142027A (en) Lithium iron phosphate battery state-of-charge monitoring and early warning method based on neural network
Lin et al. State of charge estimation with the adaptive unscented Kalman filter based on an accurate equivalent circuit model
Dai et al. ANFIS (adaptive neuro-fuzzy inference system) based online SOC (State of Charge) correction considering cell divergence for the EV (electric vehicle) traction batteries
Eddahech et al. Adaptive voltage estimation for EV Li-ion cell based on artificial neural networks state-of-charge meter
Jiang et al. State of health estimation for lithium-ion battery using empirical degradation and error compensation models
CN112147530B (en) Battery state evaluation method and device
Che et al. SOC and SOH identification method of li-ion battery based on SWPSO-DRNN
He et al. State of charge estimation by finite difference extended Kalman filter with HPPC parameters identification
Li et al. The lithium-ion battery state-of-charge estimation using random forest regression
Li et al. A novel state estimation approach based on adaptive unscented Kalman filter for electric vehicles
Samadani et al. A review study of methods for lithium-ion battery health monitoring and remaining life estimation in hybrid electric vehicles
Pei et al. An equivalent circuit model for lithium battery of electric vehicle considering self-healing characteristic
Bamati et al. Lithium-ion batteries long horizon health prognostic using machine learning
CN109839599A (en) Lithium ion battery SOC estimation method based on second order EKF algorithm
CN112083346B (en) LSTM-based parallel battery pack internal current distribution estimation method and system
Zhu et al. The SOH estimation of LiFePO4 battery based on internal resistance with Grey Markov Chain
CN104931893B (en) Suitable for the significantly extensive battery pack modeling method of parameter inconsistency
CN115542186A (en) Method, device, equipment and medium for evaluating state and consistency of energy storage battery
Tang et al. Semi-online parameter identification methodology for maritime power lithium batteries
CN112287597B (en) Lead-acid storage battery SOH estimation method based on VPGA-GPR algorithm
CN111562499B (en) Thermal management simulation method for lithium power battery of new energy automobile
Chen et al. A hybrid DNN-KF model for real-time SOC estimation of lithium-ion batteries under different ambient temperatures
Cai et al. D-ukf based state of health estimation for 18650 type lithium battery
Zhang et al. State of charge estimation for lithium-ion batteries using simple recurrent units and unscented kalman filter
Stighezza et al. Machine learning and impedance spectroscopy for battery state of charge evaluation

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
TA01 Transfer of patent application right

Effective date of registration: 20210728

Address after: 211106 No. 19 Chengxin Avenue, Jiangning Economic and Technological Development Zone, Nanjing City, Jiangsu Province

Applicant after: NARI NANJING CONTROL SYSTEM Co.,Ltd.

Applicant after: NARI TECHNOLOGY Co.,Ltd.

Applicant after: NARI Group Corp.

Applicant after: STATE GRID ELECTRIC POWER RESEARCH INSTITUTE Co.,Ltd.

Applicant after: STATE GRID JIANGSU ELECTRIC POWER Co.,Ltd.

Applicant after: State Grid Jiangsu Electric Power Co.,Ltd. Marketing Service Center

Applicant after: STATE GRID JIANGSU ELECTRIC POWER Co.,Ltd. SUZHOU BRANCH

Applicant after: STATE GRID JIANGSU ELECTRIC POWER COMPANY Research Institute

Applicant after: STATE GRID CORPORATION OF CHINA

Address before: 211106 Building 2, No.19 Chengxin Avenue, Technology Development Zone, Nanjing City, Jiangsu Province

Applicant before: NARI NANJING CONTROL SYSTEM Co.,Ltd.

Applicant before: NARI TECHNOLOGY Co.,Ltd.

Applicant before: NARI Group Corp.

Applicant before: STATE GRID ELECTRIC POWER RESEARCH INSTITUTE Co.,Ltd.

Applicant before: STATE GRID JIANGSU ELECTRIC POWER Co.,Ltd.

Applicant before: STATE GRID JIANGSU ELECTRIC POWER COMPANY Research Institute

Applicant before: STATE GRID CORPORATION OF CHINA

TA01 Transfer of patent application right
RJ01 Rejection of invention patent application after publication

Application publication date: 20200512

RJ01 Rejection of invention patent application after publication